ANNALS OF THE UNIVERSITY OF CRAIOVA

Dorel AIORDACHIOAIE

Abstract: Physical modelling is referred as the first representation of a process model and it is represented as a set of differential and algebraic equations. Noise added to model can improve the estimated behaviour of the process. Adding white noise to all variables is not recommended mainly because there could be variables based on the derivative of the white noise, and this is computational infeasible and physical impossible. The studied problem is to decide where is allowed to add noise from physical perspective early, at the modelling stage, and thus to avoid further numerical problems at the stage of simulation.
Keywords: Process Modelling, Object-Oriented Modelling Languages, Neutral Modelling, Noise Modelling, DAE models.

Camelia Avram, Adina Astilean, Mihai Hulea, Tiberiu Letia, Honoriu Valean

Abstract: The paper presents a hierarchical routing strategy for an Urban Driving Advisory System (UDAS), based on agent technology. UDAS assists the drivers to get the desired destination taking into account the current situation of traffic characteristics. It gives the estimated arrival time and the corresponding distance between a start and an arrival point. The necessary information is obtained from a real-time traffic control system (RTCS). The drivers could consult the advisory system using a variety of devices like mobile phones. The information given by the advisory system has the form of predefined short messages.
Keywords: multiagent system, GSM communication, routing algorithms

Ali Azarian, Vincent Brindejonc and Jean Marc Bruere

Abstract: A helpful survey on e-lerning in the Risk Management domain has been performed in the context of the Vicaria project (Leonardo da Vinci - N^o PP 118018) which was aimed at providing a "blended learning" system based on a combination of different media in order to optimize the open distance learning in risk assessment domain. The issue of the above mentioned investigation is summarized partly in this paper.
Keywords: software engineering, models, methods and tools, databases and multimedia, eLearning

Sorin Babii and Vladimir Cretu

Abstract: This paper presents a distributed simulator for neural networks – NetPar, and the results of several experiments in distributing the training phase of an artificial neural network. We developed and analyzed a distribution strategy for the back-propagation algorithm. We describe a distributing procedure of the well-known algorithm of back-propagation, and implemented this algorithm on several networks of computers, which allowed us to evaluate and analyze the performances using the results of actual experiments. We were interested in the qualitative aspects, trying to understand the factors which determine the behavior of this distributed algorithm. We tried to emphasize some specific aspects to be considered when implementing such a parallel algorithm on a set of workstations, interconnected in a local area network. Also, we investigated the possibilities to exploit the computational resources of such a set of workstations.
Keywords: neural networks, back-propagation algorithm, parallel algorithms, distributed algorithms, computers network.

Macarie Breazu, Antoniu Pitic

Abstract: Classical image compression methods are based on error measuring at entire image level only. In some areas there is an obvious need for getting an upper bound for the error at pixel level. In the paper we analyze such a near-lossless method based on LZW dictionary algorithm and evaluate the LRU solution for dealing with the moment when dictionary becomes full. The changes needed to adapt LZW to become a nearlossless method are also presented. As far as we know our approach is the first attempt to use LZW as a near-lossless method. Experimental results obtained and presented in the paper prove that the LRU solution gives better than the solutions based on dictionary freezing or clearing or on quadtree partitioning.
Keywords: image, compression, near-lossless, quadro, LZW, LRU.

Marius Brezovan

Abstract: In recent years, several proposals tried to associate object-oriented formalisms and Petri nets into a single framework which combine the expressive power of both approaches. This paper presents a Petri net formalism called Object Oriented High Level Petri Nets (OOHLPN), and the implementation of encapsulated multi-methods and multiple dispatching in this formalism. The encapsulated multi-methods and multiple dispatching are important feature of OOHLPNs, in addition to separating the inheritance and subtyping hierarchies.
Keywords: Petri nets, object-oriented Petri nets, object-oriented languages, formal methods

Marius Brezovan and Eugen Ganea

Abstract: It is well-known that both object-oriented paradigm and Petri net theory are two powerful frameworks used to specify complex systems, each of them having speci¯c advantages. In recent years, several proposals tried to associate them into a single framework which combine the expressive power of both approaches. This paper presents a Petri net formalism called Object Oriented High Level Petri Nets (OOHLPN), and its connection with object-oriented methodologies. One important feature of the OOHLPN formalism is treatment of the inheritance and subtyping notions. OOHLPNs use distinct hierarchies for subtyping and subclassing, the connection between a type hierarchy and an associated inheritance hierarchy being a conformance relation.
Keywords: Petri nets, object-oriented Petri nets, system modelling, formal methods, algebraic speci¯cations

Ioan Cautil

Abstract: The paper analyses the grammar of XML elements an presents algorithms for testing ambiguity of elements and to check the structure of XML documents versus the definition of the elements.
Keywords: XML

Dinu Câmpeanu and Andrei Câmpeanu

Abstract: Up until now, the only way to measure the sound quality of modern audio coding systems at low bit rates has been to implement elaborate listening tests with experienced human test subjects. Consequently, the idea of substituting the subjective tests by objective, computer based methods has been an ongoing focus of research and development. The result was PEAQ, an algorithm (ITU-R Recommendation BS.1387-1), that uses a number of psycho-acoustical tests combined to give a measure of the quality difference between two instances of a signal (a reference and a test signal). The paper describes a MATLAB PEAQ method implementation and the procedure used to assess it. Finally, PEAQ is used to rate the quality of operation for some very known digital audio editors and audio codecs.
Keywords: audio, quality, measurement, evaluation, perceptual, perceived, algorithm, codec.

Catalin Constantin Cerbulescu, Stefan Udristoiu, Claudia Monica Cerbulescu

Abstract: An important problem in both actual research and software development is content analyzing. The results of those researches are reflected in eliminating unimportant messages (spam-filter) and selecting the most important messages from a set. Present paper presents a content analyzer algorithm, implemented in a web-based application, among with his experimental results. According to an interest domain (defined by a set of words), the target content is analyzed. The result, a vector of floating numbers, is processed and analyzed, according to statistical methods. This approach is based on http://labs.google.com/sets to group words by importance and relevance.
Keywords: pattern recognition, algorithms, data processing, machine learning, statistical Analysis.

Dorian Cojocaru, Razvan Tanasie, Elena Barbulescu

Abstract: As a general concept, the job of the cluster analyze is the data partition into a number of groups, or clusters. Applying this partitioning operation on images, the image segmentation - a very important task for image processing - is obtained. This paper presents a fuzzy segmentation algorithm, fuzzy c-means, applied to grayscale images.
Keywords: computer vision, image processing, image segmentation, fuzzy logic.

Mirel Cosulschi

Abstract: Web pages created by human authors or dynamically generated using different scripts and data stored in a back-end database frequently contain many common mistakes. In order to obtain an acceptable result, web rendering engine component of a browser and automatic tools for extracting information from HTML pages need a preprocessing step to clean those web pages. We devised a simple algorithm whose main task is to properly close the tags and transform a page from HTML format into a XHTML format.
Keywords: bad data identification, data processing, transformations, html data, cleaning algorithm.

Mirela Danubianu

Abstract: Knowledge discovery from huge databases is one of the most actual problem. To solve it we can use an inductive database. With an inductive database the analyst performs a set of very different operations on data using a special-purpose language, powerful enough to perform all the required manipulations, such as data preprocessing, pattern discovery and post-processing. In this paper we present two query languages (MSQL and MINE RULE) that have been proposed for mining the association rules and discuss their common features and differences.
Keywords: association rules, query languages, data mining

Vera-Melinda GÁLFI, Liciniu-Alexandru KOVÁCS, ,Ioan-Cristian CHIFU-OROS and Stefan MOLDOVAN

Abstract: Tourism develops in all countries because it is a good source of income and, sometimes, a better solution than to work elswere. But in a global economy, a travel agent cannot promote tourism products and services only in a limited area. Wherever they go to or come from, tourists want to be well informed. As a consequence, it is a good starting point to visit and observe websites belonging to travel agencies from different regions/countries and find out how the Web service of the Internet helps these processes. Surprises may appear.
Keywords: comparative study, travel agency, website, webpage, spreadsheet, chart, hypotesys testing.

Eugen Ganea, Marius Brezovan, Simona Ganea

Abstract: Numerous advances have been made in developing intelligent systems, some inspired by biological networks. The paper discuses about then usefulness of neural networks, more specifically the motivations behind the development of neural networks, the outline network architectures and learning processes. We conclude with character recognition, a successful layered neural network application.
Keywords: neural network, training procedure, layered network, learning rule, perceptron, back propagation, character recognition.

Emilian Gutuleac

Abstract: In this paper we define a set of composition operations and descriptive based expressions to construction of composite Hybrid Stochastic Petri Nets (HSPN) for performance discrete-continuous modeling of computer systems. We consider the enhancements of our approach for a performance modeling of multiprocessor system.
Keywords: Descriptive expressions, discrete-continuous modeling, hybrid stochastic Petri nets, performance evaluation.

Anca Ion, Liana Stanescu and Dumitru Dan Burdescu

Abstract: In this paper we present an efficient and flexible software system as an alternative for the DRG-National application, used by the Romanian Government to finance the hospitals. This alternative software system is in fact an on-line application based on JSP technology and a MySQL database in order to replace the old application realized in MS Access 2000. This application will help the DRG National Bureau to analyze, at any time, real-time data, and even send it to the international organizations. The users can easily get accustom to this new application, because it keeps the structure and many menus from the old application.
Keywords: medical applications, management systems

Dorin Iordache

Abstract: In computer?s world is a permanently care of preserving secret an information, to hide an individual information or to preserve the author?s right over the respective data. To be complete this desideratum that in the time catches new dimensions and difficulty grades, appear new methods base in principal on the stenography principle who was been developed and improve.
Keywords: information technologies, criptography, steganalys, multimedia.

Isoc D, Pop M, Ignat A, Ionescu CM, De Keyser RMC

Abstract: Medical diagnosis using technical means has many applications, especially when the information to be processed is non-homogeneous. Technical means are both instrument or apparatus and software tools. The main outcomes are efficiency and an accurate information processing. This work is dedicated to extent the diagnosis means by integrating together with instrument, apparatus, and software tools, the high qualified knowledge basis existing usually in great university clinics. This integration is achieved so that any member of medical qualified staff working directly with the patient can access a specialized knowledge base using an expert system. This expert system correlates the problem identified by the clinician during the primary investigation and the similar exi-sting cases during a large time horizon inside of the clinic. The identified problem and the existing knowledge are associated inside of an intelligent predictive algorithm. The suggested approach is applied in the field of lung diseases where all the necessary features and problems are available. The outcome of intelligent integrating approach in the chosen application field empha-sizes that the results are interesting and economically efficient.
Keywords: medical diagnosis, intelligent system, case-based inference, integrated intelli-gent system, intelligent prediction.

Ioan Alfred Letia and Adrian Groza

Abstract: The speed of supply chains formation requires new modes of resolving disputes within hard time constraints. Also, the design of punishment policies applied to specific domains linking agents’ actions to material penalties is an open research issue. In our framework the principles of contract law are applied to set penalties: expectation damages, opportunity cost, reliance damages, and party design remedies, and they are introduced in the task dependency model (Walsh and Wellman, 2003). The trust is supported by providing arguments for each imposed penalty.
Keywords: artificial intelligence, agents, electronic contracts.

Ioan Alfred Letia and Radu Razvan Slavescu

Abstract: In this paper, we present a new approach for numerical trust evaluation, based on it's cognitive components of competence and willingness and how the willingness could be seen as a measure for the price of cooperation . We also show how this approach could be extended to cover the case of re-delegation, i.e. delegating a task from one agent to an intermediate one and then to the agent who actually does the job. We study how trust will be distributed among the nodes of a social network, especially in the situation when an agent possesses only partial information about the delegation chain topology and about the competence and willingness of the intermediate agents.
Keywords: artificial intelligence, agents, cooperation

Tiberiu Letia, Adina Astilean, Camelia Avram, Mihai Hulea, Honoriu Valean

Abstract: The urban vehicle traffic problems are presented with the most recent and relevant methods proposed to solve them. A solution based on the multiagent system that reacts continuously at the changing of the environment is developed. This adaptive distributed control system takes information from sensors and sends control signal to microcontroller to implement the phase durations. For the verification of the solution a real-time simulator that implements the relevant characteristics of the traffic was used.
Keywords: traffic control, distributed control, adaptive control, real-time systems, intelligent control.

Bogdan-Florin Marin, Axel Hunger, Stefan Werner

Abstract: We present an approach based on the concepts of emotions, tutoring agents and social positions to enhance and support the interaction between users within social learning environments. It advocates the use of artificial agent societies as a complement to human societies and assumes that agents will need to join such a society in order to realise user’s learning goals. This work is a more detailed description of the framework presented by Marin et al. (2004a).
Keywords: Artificial Intelligence, Software Engineering, Databases.

Ioan Z. MIHU, Horia V. CAPRITA

Abstract: The message-passing architectures consist of multiple computers interconnected through an interconnection network, communicating one with other by send-receive message functions and synchronized by barrier functions. At the moment there exist many libraries that provide a set of standardized functions for parallel programming like Message Passing Interface (MPI). MPI library allow the implementation of parallel algorithms on message-passing architectures. In this paper we propose two parallel sorting algorithms designed for message-passing architectures and implemented using MPI library: parallel Insertionsort and parallel Quicksort algorithms. We evaluate the performance of the proposed algorithms on three types of message passing architectures: linear array, two-dimensional mesh and hypercube. We evaluate the sorting time, the interprocesses communication time and the total processing time for each topology and we analyze the efficiency of the two parallel sorting algorithms related to the parallel system topology.
Keywords: message passing architecture, parallel programming, parallel algorithms, sorting algorithms

Adi-Cristina Mitea

Abstract: Data warehouses are special-purpose databases to support decision-making. In data warehouses indexing is becoming a common feature to accelerate data mining searches that combine multiple restrictive queries. Different types of indexes had been proposed and some of them are already implemented. This paper proposes a new kind of join index structure, which can be useful for multiple join operations. The algorithm to build this type of index structure is presented. Some queries, which benefit from this kind of join index, are also presented.
Keywords: indexing techniques, data warehouse, multiple join operations, relational databases.

Paulina Mitrea, Ovidiu Buza, Delia Mitrea

Abstract: Our paper presents an eLearning portal generator environment with powerful MultiMedia facilities, consisting in: advanced multimedia objects handling techniques, synchronization procedures, graphic object oriented data structures, image processing elements (based on texture analysis), moving objects extraction from the video frames, etc. First we present the general design of the e-learning site generator environment and the general scheme of the major .jsp pages, after being detailed some multimedia object handling techniques, as well as the mathematical background used in the synchronization between the moving images and the audio-video sequences.
Keywords: eLearning portal, MultiMedia technology, broadband connection, adaptive algorithm, embedded objects, synchronization, script file, tree datastructure, control kernel, histogram.

Delia Mitrea, Sergiu Nedevschi, Bogdan Fratila, Monica Lupsor

Abstract: In this paper, our purpose is to do accurate analysis and recognition of ultrasound liver images, in order to identify diffuse liver diseases like steatosis, cirrhosis and hepatitis. In order to do a proper tissue analysis from ultrasound images, we compare the efficiency of texture-based methods like the Gray Level Cooccurrence Matrices (GLCM), fractals and the texton-based method.
Keywords: ultrasound liver images, biomedical image recognition, decision making, accuracy, statistics, texture, GLCM, fractals, textons

Mihai Mocanu, Mihai Dorobantu and Sorin Capra

Abstract: Network traffic analysis is employed in order to provide continuously a communication map for all the computers in a local network, necessary to network administrators, analysts, developers a.s.o. In order to get detailed information any network must be monitored. Library calls from a component added to the OS, called WinPcap, can be used to capture network packets and to generate traffic. Traffic generation can be then used to test software components (capture applications) or hardware components (network adapters, modems) in a large variety of tools (applications) specifically designed for network analysis, troubleshooting, security and/ or monitoring. Design principles for an efficient, multithreaded tool to do traffic analysis and sniffing in an Ethernet LAN are presented in the final part of the paper.
Keywords: packet capture, network analysis, libpcap, Winpcap

Cristina Consuela Petre

Abstract: Lightweight web-based diagramming tools are designed to provide easy and rapid production of diagrams. The use of the web brought up advantages as high portability and usability and the possibility of collaborative work. Although thin-client design tools lack problems like installation or setup overhead, difficult to understand user interfaces, building them is a challenging task. This paper evaluates the idea of realizing an extension to a thick-client tool which enables any specified diagram designer to be realized as a thin-client tool. The thin-client applications are accomplished by means of server side components that interact with the thick-client designer to produce GIF diagrams for display and editing in the user’s web browser. An analysis of the effectiveness of the present approach is made and potential future development is identified.
Keywords: Thin-client, thick-client, web-based diagramming tool, web-browser

Serban B. PETRESCU, Tudor C. METEA, Catalin F. TUDOSE

Abstract: Threedimensional reconstruction of geographic zones relief is a process with wide applications in GIS type products, especially those that implement models for simulation of various phenomenons that may affect a specific geographic area (floods, terrain slips etc). By threedimensional reconstruction of geographic area relief starting from serial sections, we understand regenerating of the area surface in the form of a set of connected triangles. The results of the reconstruction process may be introduced in variuos threedimensional vizualization programs, or used as an input for relief analysis programs.
Keywords: Three-dimensional modelling, serial sections, level curves, polygon overlapping, tiling, triangularization, simple branching

Teo Popa and Mihai Mocanu

Abstract: Medical imaging plays a crucial role in the diagnosis techniques upon which modern medical treatment depends. In concert with the increasing number of imaging techniques based on different physical principles and the availability of relatively inexpensive computational resources, more and more medical software analysis tools are developed. We present in this article an experimental visualization system called MEVIAN (MEdical VIsualization & ANalysis) that we believe it may help the radiologist in the diagnostic of liver pathology. This system is based on VTK, ITK and FLTK open source libraries that are freely available for download and provides the vital components for diagnosis of liver lesions and a reliable differential diagnosis. Using this tool the radiologist can perform accurate preoperative localization of the focal or diffuse lesion, both segmental and topographical.
Keywords: medical imaging, open source, VTK, ITK, FLTK.

Resceanu Ionut

Abstract: E-learning is the continuous assimilation of knowledge and skills by adults stimulated by synchronous and asynchronous learning events — and sometimes Knowledge Management outputs — which are authored, delivered, engaged with, supported, and administered using Internet technologies. The first e-learning systems were actualy web sites which contained some electronic materials and primitive tests. Also, there were few categories of actors within those systems. Modern age e-learning systems are based on the notion of lesson. The people involved are grouped within several categories. The present paper tries to present those categories and also to highlight the structure of the main entity of modern e-learning systems – the lesson.
Keywords: learning systems, computer applications, computer systems, computer-aided instruction

Rosu Marius, Phan Cong Chinh, Stefan Freinatis

Abstract: Today's expert systems deal with domains of narrow specialization and the set of observable facts is limited. When trying to enlarge the specialization domain, the limited number of facts becomes inaccurate. This paper presents an attempt to solve the inaccuracy problem of wide domain expert system by modeling the domain of knowledge and formalizing the reasoning process. The intelligence of the systems is used for selection of possible observable facts as well as for the prediction or diagnostics function of the expert system.
Keywords: rule based expert system, ontology, intelligent dialogue.

Cosmin Stoica Spahiu, Costin Badica, Gabriel Vladut and Michael J. Roberts

Abstract: This article presents an original implementation of an e-commerce supermarket system – SUM. To be competitive in today’s world, manufacturers need to enrich the range of products they can sell in a make-to-order environment, to minimize the delivery time to the client and to maximize the production in order to improve profitability. The best solution is to eliminate the retailers and to sell directly to the clients, using specialized software that groups the received orders in batches, on family of products. The paper briefly summarizes the software architecture of the SUM system, the concepts and impact of the software.
Keywords: e-commerce, make-to-order, supply chain, management, software architecture

Tanasie Razvan, Tunaru Cristina

Abstract: This paper focuses on animation creation based on time dependent transformations in a real case and presents a solution that uses computer graphics techniques in interactive applications. The application is elaborated starting with a story theme and every character or element of the scene has a transformation matrix attached that defines the final movement of the object.
Keywords: computer graphics, rendering, DirectX, mesh, vision.

Philippe Trigano, Elvira Popescu, Ecaterina Giacomini

Abstract: Since the project-based learning approach is advocated in the literature as one of the most effective methods of learning, our goal is to design a series of pedagogical scenarios involving project developing. IMS Learning Design specification proves very effective for this task, providing facilities for both individual project-based learning as well as for collaborative work, in case of projects done in groups. Our research is accomplished in the context of netUniversité e-learning platform, which integrates the functionality of both a Learning Management System and a Learning Content Management System. This paper covers the process of modelling two project-based pedagogical scenarios using IMS LD and their implementation in the netUniversité platform.
Keywords: project-based learning, collaborative work, instructional design, IMS Learning Design specification, educational hypermedia.

Zmaranda Doina, Rusu Claudia, Gligor Marius

Abstract: In this paper we present a tool for modelling and analyzing real-time systems. The description of the system being modelled is based on system’s timing characteristics. Using the implemented tool, several quantitative timing information about the system such as schedulability, response time and loading factor could be estimated. Based on the developed simulated model of a real-time system, the timing behavior could be assessed prior to implementation.
Keywords: Real-time system, deadline, schedulability, Worst Case Execution Time (WCET).